A Highly Available Cloud-Based Real Estate Marketplace Using Elastic Load Balancing and Multi-Zone Redundancy

Abstract: Online property marketplaces have transformed how real estate is discovered, listed, and transacted, yet the platforms that host them must withstand highly variable traffic, sustain rich media catalogues, and remain continuously available, since downtime during peak demand directly translates into lost opportunities. Conventional single-server deployments are vulnerable to overload and represent a single point of failure, while naive scaling strategies inflate cost without guaranteeing resilience. This paper presents a cloud-based real estate marketplace engineered for high availability through elastic load balancing and redundant, multi-zone deployment. The platform combines a Java service backend with a Node.js web client and distributes incoming requests across automatically scaled application instances spanning two availability zones, with a managed relational database replicated for failover and an object store and cache supporting media-rich listings. Health-checked load balancing reroutes traffic away from unhealthy instances, while auto-scaling adjusts capacity to demand. Experimental evaluation under simulated load demonstrated an average response time of 110 milliseconds at one thousand concurrent users and graceful degradation up to ten thousand, alongside a 99.98% availability score, substantially outperforming a single-server baseline whose latency and failure rate rose sharply. The principal contributions of this work are a resilient, load-balanced, multi-zone marketplace architecture, an integrated failover and auto-scaling strategy that sustains continuous service, and an empirical demonstration of superior availability, scalability, and throughput relative to conventional deployments.

Keywords: Real estate marketplace, high availability, elastic load balancing, cloud computing, auto-scaling, fault tolerance, multi-zone redundancy, web application.